model releases

HiP-LoRA: Budgeted Spectral Plasticity for Robust Low-Rank Adaptation

arXiv:2604.17751v1 Announce Type: cross Abstract: Adapting foundation models under resource budgets relies heavily on Parameter-Efficient Fine-Tuning (PEFT), with LoRA being a standard modular solutio

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

HiRAS: A Hierarchical Multi-Agent Framework for Paper-to-Code Generation and Execution

DGX agent

arXiv:2604.17745v1 Announce Type: new Abstract: Recent advances in large language models have highlighted their potential to automate computational research, particularly reproducing experimental resu

model-releasesarxiv-cs-cl

21 Apr 2026

HORIZON: A Benchmark for In-the-wild User Behaviour Modeling

arXiv:2604.17259v1 Announce Type: cross Abstract: User behavior in the real world is diverse, cross-domain, and spans long time horizons. Existing user modeling benchmarks however remain narrow, focus

HorizonBench: Long-Horizon Personalization with Evolving Preferences

arXiv:2604.17283v1 Announce Type: new Abstract: User preferences evolve across months of interaction, and tracking them requires inferring when a stated preference has been changed by a subsequent lif

How Robustly do LLMs Understand Execution Semantics?

arXiv:2604.16320v1 Announce Type: cross Abstract: LLMs demonstrate remarkable reasoning capabilities, yet whether they utilize internal world models or rely on sophisticated pattern matching remains o

How Should We Enhance the Safety of Large Reasoning Models: An Empirical Study

arXiv:2505.15404v2 Announce Type: replace Abstract: Large Reasoning Models (LRMs) have achieved remarkable success on reasoning-intensive tasks such as mathematics and programming. However, their enha

HQA-VLAttack: Towards High Quality Adversarial Attack on Vision-Language Pre-Trained Models

arXiv:2604.16499v1 Announce Type: new Abstract: Black-box adversarial attack on vision-language pre-trained models is a practical and challenging task, as text and image perturbations need to be consi

HUGGING FACE JUST AUTOMATED THEIR ENTIRE POST-TRAINING TEAM WITH AN AGENT. It reads papers, runs GPU experiments, iterates, and builds resea…

model-releasesclem-delangue--x

HUGGING FACE JUST AUTOMATED THEIR ENTIRE POST-TRAINING TEAM WITH AN AGENT. It reads papers, runs GPU experiments, iterates, and builds research-backed models autonomously. Pushed a benchmark from 10%

Hybrid-Vector Retrieval for Visually Rich Documents: Combining Single-Vector Efficiency and Multi-Vector Accuracy

arXiv:2510.22215v2 Announce Type: replace-cross Abstract: Retrieval over visually rich documents is essential for tasks such as legal discovery, scientific search, and enterprise knowledge management.

I agree, and I will add to this. Companies should own their memories, skills and other resources agents need. Use Claude code, codex, copilo…

model-releasesharrison-chase--x

I agree, and I will add to this. Companies should own their memories, skills and other resources agents need. Use Claude code, codex, copilot, etc for intelligence but your execution layer should plug

I came up with a somewhat foolish new benchmark for testing image generation models, to exercise the new ChatGPT Images 2.0: 'Do a where's W…

model-releasessimon-willison--x

I came up with a somewhat foolish new benchmark for testing image generation models, to exercise the new ChatGPT Images 2.0: 'Do a where's Waldo style image but it's where is the raccoon holding a ham

I find that open weights models over-perform on benchmarks compared to actual real-world usage, and Kimi feels like no exception. For exampl…

model-releasesethan-mollick--x

I find that open weights models over-perform on benchmarks compared to actual real-world usage, and Kimi feels like no exception. For example, a small amount of use will show that Kimi is not as good

I hope this helps. If you need more help let me know: Connecting OpenClaw to the X API is straightforward now thanks to X’s official native …

model-releaseselon-musk--x

I hope this helps. If you need more help let me know: Connecting OpenClaw to the X API is straightforward now thanks to X’s official native support... The best and most direct method uses the official

ICAT: Incident-Case-Grounded Adaptive Testing for Physical-Risk Prediction in Embodied World Models

arXiv:2604.16405v1 Announce Type: cross Abstract: Video-generative world models are increasingly used as neural simulators for embodied planning and policy learning, yet their ability to predict physi

IDOBE: Infectious Disease Outbreak forecasting Benchmark Ecosystem

arXiv:2604.18521v1 Announce Type: new Abstract: Epidemic forecasting has become an integral part of real-time infectious disease outbreak response. While collaborative ensembles composed of statistica

iDocV2: Leveraging Self-Supervision and Open-Set Detection for Improving Pattern Spotting in Historical Documents

arXiv:2604.16726v1 Announce Type: new Abstract: Considering the imminent massification of digital books, it has become critical to facilitate searching collections through graphical patterns. Current

Improving LLM Code Reasoning via Semantic Equivalence Self-Play with Formal Verification

arXiv:2604.17010v1 Announce Type: new Abstract: We introduce a self-play framework for semantic equivalence in Haskell, utilizing formal verification to guide adversarial training between a generator

In-Context Symbolic Regression for Robustness-Improved Kolmogorov-Arnold Networks

arXiv:2603.15250v2 Announce Type: replace Abstract: Symbolic regression aims to replace black-box predictors with concise analytical expressions that can be inspected and validated in scientific machi

Incoherent Deformation, Not Capacity: Diagnosing and Mitigating Overfitting in Dynamic Gaussian Splatting

arXiv:2604.16747v1 Announce Type: new Abstract: Dynamic 3D Gaussian Splatting methods achieve strong training-view PSNR on monocular video but generalize poorly: on the D-NeRF benchmark we measure an

IncreFA: Breaking the Static Wall of Generative Model Attribution

arXiv:2604.17736v1 Announce Type: new Abstract: As AI generative models evolve at unprecedented speed, image attribution has become a moving target. New diffusion, adversarial and autoregressive gener

Inflated Excellence or True Performance? Rethinking Medical Diagnostic Benchmarks with Dynamic Evaluation

arXiv:2510.09275v2 Announce Type: replace Abstract: Medical diagnostics is a high-stakes and complex domain that is critical to patient care. However, current evaluations of large language models (LLM

Injecting Structured Biomedical Knowledge into Language Models: Continual Pretraining vs. GraphRAG

arXiv:2604.16422v1 Announce Type: new Abstract: The injection of domain-specific knowledge is crucial for adapting language models (LMs) to specialized fields such as biomedicine. While most current a

INTENT: Invariance and Discrimination-aware Noise Mitigation for Robust Composed Image Retrieval

arXiv:2604.18051v1 Announce Type: new Abstract: Composed Image Retrieval (CIR) is a challenging image retrieval paradigm that enables to retrieve target images based on multimodal queries consisting o

InternScenes: A Large-scale Simulatable Indoor Scene Dataset with Realistic Layouts

arXiv:2509.10813v3 Announce Type: replace Abstract: The advancement of Embodied AI heavily relies on large-scale, simulatable 3D scene datasets characterized by scene diversity and realistic layouts.

Interpolating Discrete Diffusion Models with Controllable Resampling

arXiv:2604.17310v1 Announce Type: new Abstract: Discrete diffusion models form a powerful class of generative models across diverse domains, including text and graphs. However, existing approaches fac

Introducing ChatGPT Images 2.0

ChatGPT Images 2.0 is an updated version of OpenAI's image generation feature integrated into ChatGPT, offering improvements to image creation capabilities for users. The update likely includes enhanc

model-releasesopenai

Introducing ChatGPT Images 2.0 A state-of-the-art image model that can take on complex visual tasks and produce precise, immediately usable …

Introducing ChatGPT Images 2.0 A state-of-the-art image model that can take on complex visual tasks and produce precise, immediately usable visuals, with sharper editing, richer layouts, and thinking-

model-releasesopenai--x

Introducing ml-intern, the agent that just automated the post-training team @huggingface It's an open-source implementation of the real rese…

model-releasesclem-delangue--x

Introducing ml-intern, the agent that just automated the post-training team @huggingface It's an open-source implementation of the real research loop that our ML researchers do every day. You give it

Judge a Book by its Cover: Investigating Multi-Modal LLMs for Multi-Page Handwritten Document Transcription

arXiv:2502.20295v2 Announce Type: replace-cross Abstract: Handwriting text recognition (HTR) remains a challenging task. Existing approaches require fine-tuning on labeled data, which is impractical t

JudgeMeNot: Personalizing Large Language Models to Emulate Judicial Reasoning in Hebrew

arXiv:2604.18041v1 Announce Type: new Abstract: Despite significant advances in large language models, personalizing them for individual decision-makers remains an open problem. Here, we introduce a s

Jupiter-N Technical Report

arXiv:2604.17429v1 Announce Type: new Abstract: We present Jupiter-N, a hybrid reasoning model post-trained from Nemotron 3 Super, a fully open-source 120 billion parameter LLM. We target three object

K2.6 + hermes = 4 hr session setting up qwen 3.6 training regime on dgx spark with the current autnomous session currently lasting 70+ min w…

model-releasesnous-research--x

K2.6 + hermes = 4 hr session setting up qwen 3.6 training regime on dgx spark with the current autnomous session currently lasting 70+ min without any prompting. Kimi with hermes is next level. @NousR

Keep improving 🚀🚀 @arena

Keep improving 🚀🚀 @arena Qwen3.6 Plus lands at #7 in Code Arena with a score of 1476 - up +16 points since the Preview. The new score also moves @AlibabaGroup to #3 lab in Code Arena. In the Text Aren

model-releasesqwen--x

📢 Kimi K2.6 API is live • Input Price (Cache Hit): 0.16 / M tokens • Input Price (Cache Miss): 0.95 / M tokens • Output: $4.00 / M tokens…

📢 Kimi K2.6 API is live • Input Price (Cache Hit): 0.16 / M tokens • Input Price (Cache Miss): 0.95 / M tokens • Output: $4.00 / M tokens Kimi K2.6 is our latest + most intelligent model - stronger lo

Kimi K2.6 autonomously overhauled exchange-core, an 8-year-old open-source financial matching engine. Over a 13-hour execution, the model it…

Kimi K2.6 autonomously overhauled exchange-core, an 8-year-old open-source financial matching engine. Over a 13-hour execution, the model iterated through 12 optimization strategies, initiating over 1

Kimi K2.6 has captured #1 on the open-weight Vals Index, and is #7 overall.

Kimi K2.6, a language model developed by Moonshot AI, has achieved the top ranking on the open-weight category of the Vals Index benchmark, while placing 7th overall across all model categories. The V

Kimi K2.6 is now live inside Anything!

Kimi K2.6, an AI model from Moonshot, has been integrated into the Anything platform. This update likely enables users to access Kimi's capabilities directly within the Anything application interface.

Kimi K2.6 @Kimi_Moonshot is the new leading open-weights agent model, landing at #4 on Claw-Eval (Pass^3: 62.3%). Key takeaways: - 👑 Best o…

Kimi K2.6 @Kimi_Moonshot is the new leading open-weights agent model, landing at #4 on Claw-Eval (Pass^3: 62.3%). Key takeaways: - 👑 Best open-source agent, period: Pass^3 of 62.3% is the highest of a

KIRA: Knowledge-Intensive Image Retrieval and Reasoning Architecture for Specialized Visual Domains

arXiv:2604.16915v1 Announce Type: new Abstract: Retrieval augmented generation (RAG) has transformed text based question answering, yet its extension to visual domains remains hindered by fundamental

Knowing When to Quit: A Principled Framework for Dynamic Abstention in LLM Reasoning

arXiv:2604.18419v1 Announce Type: cross Abstract: Large language models (LLMs) using chain-of-thought reasoning often waste substantial compute by producing long, incorrect responses. Abstention can m

Knowledge without Wisdom: Measuring Misalignment between LLMs and Intended Impact

arXiv:2603.00883v2 Announce Type: replace Abstract: LLMs increasingly excel on AI benchmarks, but doing so does not guarantee validity for downstream tasks. This study contrasts LLM alignment on bench

Language Models Don't Know What You Want: Evaluating Personalization in Deep Research Needs Real Users

arXiv:2603.16120v2 Announce Type: replace Abstract: Deep Research (DR) systems help researchers cope with ballooning publishing counts. Such tools synthesize scientific papers to answer research queri

Large Language Models Are Still Misled by Simple Bias Ensembles

arXiv:2505.16522v3 Announce Type: replace Abstract: With the evolution of large language models (LLMs), their robustness against individual simple biases has been enhanced. However, we observe that th

Late Fusion Neural Operators for Extrapolation Across Parameter Space in Partial Differential Equations

arXiv:2604.16721v1 Announce Type: new Abstract: Developing neural operators that accurately predict the behavior of systems governed by partial differential equations (PDEs) across unseen parameter re

Latent Preference Modeling for Cross-Session Personalized Tool Calling

arXiv:2604.17886v1 Announce Type: new Abstract: Users often omit essential details in their requests to LLM-based agents, resulting in under-specified inputs for tool use. This poses a fundamental cha

LayerCache: Exploiting Layer-wise Velocity Heterogeneity for Efficient Flow Matching Inference

arXiv:2604.16492v1 Announce Type: new Abstract: Flow Matching models achieve state-of-the-art image generation quality but incur substantial inference cost due to iterative denoising through large Tra

LEAF: Knowledge Distillation of Text Embedding Models with Teacher-Aligned Representations

arXiv:2509.12539v2 Announce Type: replace-cross Abstract: We present LEAF ('Lightweight Embedding Alignment Framework'), a knowledge distillation framework for text embedding models. A key distinguish

Learned Nonlocal Feature Matching and Filtering for RAW Image Denoising

arXiv:2604.17453v1 Announce Type: cross Abstract: Being one of the oldest and most basic problems in image processing, image denoising has seen a resurgence spurred by rapid advances in deep learning.